The sciences of the artificial (3rd ed.)
The sciences of the artificial (3rd ed.)
The Design and Analysis of Computer Algorithms
The Design and Analysis of Computer Algorithms
Human Problem Solving
Semantic Information Processing
Semantic Information Processing
| Hi-index | 0.00 |
It is conjectured that a good cognitive psychology theory will lead to a good artificial intelligence (AI) program. If this is true there should be a convergence of psychological and AI considerations in theory construction. This convergence is illustrated in terms of ACT, a computer simulation model of cognitive processes. Separate AI and psychological considerations are used to motivate the decision to design ACT as a production system operating on an network data base. Similar motivation is provided for other features of ACT implemented within this framework. These features include the use of a propositional structure for the associative network, a spreading activation process operating on the network, the simulated ability to execute several procedures in parallel, and the use of strength measures to select among competing productions and competing paths in the network.We have been working on a production system model of human cognition called ACT. An earlier version of the ACT system, called ACTE, is described in Anderson [3], Anderson, Kline, and Lewis [5], and Kline and Anderson [2]. That system has been used to develop mini-models for retrieval from memory, inference making, language comprehension, question-answering, and problem solving. We are currently working on a new version of ACT called ACTF. This paper discusses a number of the design decisions underlying the ACT system. We will discuss how these design decisions are motivated by both psychological and artificial intelligence (AI) considerations.